Thread 106245512

Why can Wayland compositors still not match the cursor responsiveness and smoothness of Xorg? Even something as simple as navigating the GNOME shell on the newest version of OpenSUSE Leap causes it to skip. There was also that recently published article that showed a slight latency edge on Xorg with a slow-motion camera.

Well, you could read the readily available bug trackers and technical discussions, but since you're asking on /g/, I assume you want an explanation that blames transsexuals and Indians.

>>106245512 (OP)
That comes from fundamenal design difference. This will never be fixed, because Wayland forces composition which is optional on X. X support software cursor and Wayland forces everything to be dependent on GPU. RIP low end old usable PCs btw. I think that Wayland is also pushed for planned obsolescence reasons so people dump their old usable computers and give money to companies that can't innovate anymore and need "constant growth".

Wayland devs are so obsessed with their muh perfect frames that the possibility of the cursor tearing is unbearable to them. So they make it lag instead. It's by design.

pure retardation.

supposedly they made Wayland to fix latency and performance and what we got is something that performs worse especially when you use NVIDIA

>>106245912
Neither Xorg nor Wayland force either hardware or software cursors. Nobody sane uses software cursors anyway.

>>106245983
The cursor never tears on Xorg. It's a non-issue.

>>106245512 (OP)
gayland basically wants every frame to be "perfect", so it forces vsync even on the mouse cursor. x11 doesn't do that. in theory x11 can and will introduce tearing because of that but TearFree exists and it works, so

This has nothing to do with muh perfect frames or muh compositing.

The real answer is that X.org creates 1000 syscalls per second whenever you move the mouse. This is effective, but highly inefficient. There's no reason to update the mouse any faster than what your display can achieve, unless you need a second heater in your home.

>>106245512 (OP)
>Even something as simple as navigating the GNOME shell on the newest version of OpenSUSE Leap causes it to skip
That's just gnome retardation

>>106245512 (OP)
>gnome
Works just fine for me on KDE

>>106246339
yet Xorg performs better on old machines

>>106246465
on new machines too desu

>>106245512 (OP)
It also manages to be worse than just regular composited Xorg so it's pointless to use.

>>106246291
>Nobody sane uses software cursors anyway.
this is hilarious because i'm viewing this very thread using x11vnc utilizing the TightVNC CursorPosUpdates and CursorShapeUpdates extensions.
nta so you don't reply like a retard.

>>106245512 (OP)
what the absolute fuck are you deranged X trannies even talking about now?
what is with this insane cope? every day is a new goal post for the X tranny.

>>106246291
>Nobody sane uses software cursors anyway.
I'm using software cursor because hardware cursor doesn't behave well with VRR

>>106246291
On Wayland the cursor plane is synchronized with the primary plane. It's not a software cursor but it behaves identically.

>>106246465
>>106246700
Only if you're using shitty proprietary drivers. Just running eglgears raises my CPU to 5% on xfwm4 (other X compositors are even worse). It's measurably lower on labwc.

>>106247163
GNOME stopped doing since at least 45.
https://www.phoronix.com/news/GNOME-Mutter-KMS-Thread
I believe KDE also does something similar. I can definitely feel a difference between Debian 12 and Debian 13. I actually liked the old behavior more as the cursor was perfectly synchronized with everything else.

>>106247417
>labwc
>wlroots
>no nvidia support
opinion discarded use case not found

>>106245615
>I assume you want an explanation that blames transsexuals and Indians.
Yes, that would do nicely, thank you

>>106247417
It's possible to synchronize the cursor only while dragging things but then the latency is inconsistent. It's always a compromise.

>>106247447
Yep, called it with the drivers. I rest my case.

>>106247461
So what Windows does? No, thanks.

>>106247417
KDE has proper scheduling and commits cursors updates three times during refresh cycle
https://zamundaaa.github.io/wayland/2023/08/29/getting-rid-of-cursor-stutter.html

>>106246006
>buys a GPU from jewish manufacturer that won't release driver sources
>surprised it runs like shit when everything must be reverse-engineered
Wayland should just drop all nNvidia cards so you faggots are forced back to Windows or to buy GPU from manufacturer that actually cares about Linux.

>>106247582
>Wayland should just drop all nNvidia cards s
and be more irrelevant yes go for it

>>106245512 (OP)
>Why can Wayland compositors still not match the cursor responsiveness and smoothness of Xorg?
Had to upgrade to something using Wayland to even get a smooth cursor on my machine(tm). The lowest-end nVidia Quadro didn't get smooth cursors on X with nouveau (and the official driver broke a bunch of random other shit). Now I use the open source AMD driver with an RX 550 instead, which works well, especially with Wayland.

>>106246006
>supposedly they made Wayland to fix latency and performance
And it is supposed to be more secure and leave less room for security exploits and bugs, first and foremost. Not that anyone cares. And wayland sucks anyway.

>>106247591
based accelerationist

>>106247082
>muh trannies
OBSESSED
Take your disgusting fetish to /b/ or /d/. Now shoo.

>>106247532
>As a result of that, I consider fixing the kernel in this regard as just not possible at the moment.

It's been two years now. Has that changed?

>>106247163
I find this entire thread of discussion offensive to software cursors. Software cursors are not even laggy. They can be just as responsive as hardware cursors if done right.

>>106247082
>X is for trannies
>Wayland is for trannies
The rightoids are so confused right now

>>106247889
the rightoids use xlibre

>>106247917
apparently the rightoids haven't seen the featured screenshot on the GitHub.

>>106247933
maybe the tranny is a rightoid too

>>106247933
>What is a joke

>>106247941
>>106247950
Awh :3c

>>106247933
>left drone: no no you making fun of us we wont promote it
>right drone: look the support transsexuals i don't understand sarcasm
drones being drones

>>106247933
>Wayland is gay
>Xorg is trans
Soo, Wayland is the lesser evil. Got it. Never touching Xtrans again!

>wayland/xorg thread
>devolves into bait
was surprisingly fast this time, too.

>>106248025
I don't know enough about either so I'm trying to see which one the retards gravitate to.
I just install stock fedora + gnome and it works well enough. All of my real work happens on windows 11 anyway.

>>106247889
Every single thing on Linux is gay and trans, on account of it being created by gay transsexual.

>>106248060
one doesn't tear, the other does.
one has more than 8bit color, HDR, independent monitor refresh and scale, the other doesn't.
one actually has a future, the other doesn't.

it's that simple. if you're still holding onto X, you're paying a performance penalty at best moving forward.

>kde wayland for desktop programs
>uncomposited fluxbox on another tty for games
just works.

works on my machine

Yeah but Gayland supports HDR and xorg doesn't.

>>106248191
>HDR
when ever y enable HDR on my monitor it burns my eyes is just me or HDR needs buring eye tech to even be visible?

>>106245512 (OP)
It's not only the input latency (which is the highest across all known display systems people use for now). If you try to lower the scrolling speed, you'll notice it doesn't actually make each scroll slower. It just ignores some of the wheel's movement, while the rest of the movement is of the same speed as before. If you set the speed to 0.5 for example, half of the scroll wheel movement gets ignored. I don't know why it was designed that way, but it's weird. The main input device on a desktop environment feels quite neglected desu. The way mice work on wayland implementations is the main reason I can't use any of them for long. I turn back to xorg because my hand gets tired.

>>106245615
I’d rather blame Guatemalans like Harry Potter

>>106246291
I use a software cursor because it feels more responsive than hardware cursor. Amd driver also has some weird ass bug that causes the cursor to lag if I dont use software cursor

>>106249141
>decades of windowing system development
>the trannies still manage to fuck up cursor
grim

>>106249141
how do i know if im using hardware/software cursor

>>106249181
You can run the drm_info program when the cursor is visible. Search for "= Cursor" in the output to find the cursor planes. If there is one with FB_ID set to a non-0 value then you have hardware cursor enabled.
This drm_info only works on amd/intel x11 or wayland, not x11 nvidia (since it doesn't use linux drm api)

>>106249211
~: drm_info | grep Cursor
drmModeGetResources: Operation not supported
Failed to retrieve information from /dev/dri/card1
───"type" (immutable): enum {Overlay, Primary, Cursor} = Primary


:( im on nvidia

GNOME 48.

>>106250239
You're showing the latency of the compositor for the window, not the cursor

>>106248153
>one doesn't tear, the other does.
Xorg server got a fix for that 3 years ago and it was enabled by default but it was never put in a release but xorg development is run by wayland devs that have stated that they want to kill x11. If you run the latest xorg server then you wont have tearing.
>one has more than 8bit color
So does x11, startx -- -depth 30
>independent monitor refresh and scale
So does x11: https://streamable.com/69x7et

>>106245512 (OP)
Wayland is hot garbage, nobody wanted it and now that the Wayland people aren't in control of X any more it will die because they can't continue to kill X.

>>106250770
That's the point, all compositors keep the rendered content and cursor plane in sync even though they can be independent. This fixes the issue in
>>106250239
as can be seen in
>>106250250
Of course, now you are tied to the frame clock and it feels unresponsive

>>106248298
In mine it doesn't but all colors look overcooked in games, so not really worth it to enable it.
Wasn't a problem with my LG C3 tv, but this supposedly top notch msi qd-oled has been utter trash since I got it.

>>106245615
Unprompted self report

>>106247582
>manufacturer that actually cares about Linux.
Like who

>Use my low-end laptop
>Disable the dedicated input thread with -dumbSched
>Increase nice value
>Run shadertoy and stress -c 4
>Cursor still never hitches
What is this black magic? How the fuck does Xorg do it?

>>106245512 (OP)
True, on wayland (latest kde plasma) my cursor lags as much as the game im playing, so it gets tied to the game fps. On x11 my cursor is always a smooth 144fps like my monitor.

>>106251510
>InputThread

>>106251626
Picture somewhat unrelated.

>>106248324
There's no fucking way that's real.

>>106247889
>rightoids
It's over a decade later and /leftypol/ and friends still can't meme.

Wayland compositors are not that bad, I don't have anything bad to say about Xorg either, just werks I guess. However what pisses me off is the fact that XFCE and other less funded DEs are even thinking of implementing their own Wayland compositors instead of sticking with an already out of the box solution like labwc (like they are doing for now). I swear "not invented here syndrome" is gonna kill all these projects in this Wayland push.

>>106245512 (OP)
Frankly told becauae they do it for a job.

>>106252039
>not invented here syndrome
Except that's not exactly what's happening. Xfwm is being based on wlroots just like labwc. It's not reinventing the wheel.
Besides, labwc is just a shitty Openbox clone. It's fit for LXQt, but not for XFCE.

>>106252555
>wlroots
isn't wlroots unmaintained hot turd written in C that has tons of silent bugs?

>>106250802
They also deleted a bunch of old but perfectly fine drivers "just because lel" and those are going back into XLibre.

>>106252622
>Unmaintained
Not even close. The last two releases are two months apart IIRC.
>Hot turd
It's extremely lightweight
>Written in C
And that's a good thing.
>tons of silent bugs?
It hasn't frozen up on me like Hyprland has, so I guess that passes as silent.

>>106252816
>posted it again
>even though it proves OP right
>again

>>106252825
Why would you assume that I posted it with the intent of refuting his point.